The ABCs of the "New Normal" for Streaming Services
Analytics. Bandwidth. Compression. These three words have always been important to the success of streaming services, and in today's world, they've never been more vital. Around the globe, spikes in streaming TV viewing continue to be exponential. In fact, 87% of U.S. consumers say they've consumed more content during the COVID-19 pandemic, according to a Global Web Index report. And yet, as stay-at-home restrictions continue to loosen, streaming TV viewership remains higher than pre-pandemic times. With this consistent demand from viewers for content, the time is right for video service providers to leverage the ABCs (analytics, bandwidth and compression) for not just short-term needs, but longer term, sustained growth.
Thankfully, both technology and expertise are now more available than ever before to help address such hurdles, particularly related to visibility and control capabilities that help video service providers more efficiently and effectively serve their subscribers. Furthermore, end-to-end expertise can seamlessly enable the analytics, bandwidth, and compression required to help maintain and grow subscriber bases.
When we consider the near-overnight viewing spikes we saw in March, we likely first think of bandwidth needs. Such concerns are not new for our industry (think of live sports viewing over streaming, for example), but since COVID-19 they have evolved and even grown. The pandemic has brought to light questions that we've been talking about for some time, but may have not completely answered, such as the following:
- How can we properly scale?
- What needs to be taken into consideration to preemptively scale? Geography? Scheduled events? Content popularity?
- How can service providers leverage artificial intelligence (AI) and analytics to help scale in advance?
- How can we alleviate the pressure of more demanding video applications on the networks?
Closely tied to bandwidth is the amount of bitrate required to deliver the high video quality viewers have come to expect. Early on during the COVID-19 pandemic, bandwidth was such a concern that many providers quickly reduced video service bitrates by percentages in the double digits. Across Canada and Europe, for example, Netflix reduced network traffic by 25% in an effort to sustain good quality in late March. In early May, there were reports that bit rates were actually less than 50% of what was the norm. It's been reported that these network reductions have caused noticeable lowered image quality, such as blurring and pixelation.
Netflix is not alone—many video services, if not all, are still trying to find the most effective route to address the bandwidth challenge. Negatively impacting quality of experience (QoE) for viewers is simply not an option—they'll just go to another service that promises better video quality. Fortunately, tools such as artificial intelligence (AI) and machine learning (ML) are making strong inroads and seeing positive results, particularly when it comes to compression techniques for maintaining the best quality at the lowest bitrate.
Next-generation video specifications can make video compression more efficient, but also require substantial encoder complexity. By using tools such as ML in a smart way, we can find a balance between bandwidth-efficient video streams and encoder complexity. AI and ML allow us to not only compress whole or partial frames individually, but also determine the areas within each frame that will matter least to viewers and compress them more than the others. By leveraging AI and ML as a tool for codec optimization, matching the high sensitivity to image quality and abrupt changes that are detected by the human eye becomes more intuitive and helps to significantly improve adaptive bitrate encoding at the live program or event level.
Content aware encoding (CAE) by scene can further help reduce the need for more bandwidth, and together with smarter quality metrics, bring us closer to constant perceptual quality. Moving so quickly across so many geographies and at such scale leads to an explosion in the computational requirements of encoders. This means that the need for faster innovations and more efficient codecs have become immediate. Streaming services looking to expand the geographies they serve can benefit greatly by tapping into the capabilities of CAE and working with a technology partner who understands the ins and outs of the various codecs, implementation approaches and scalability requirements to ensure high QoE while the expansion is underway.
When considering QoE, it's important to think of content delivery networks (CDNs). Whether for video on demand (VOD), live, or TSTV content, CDNs bring the caching process closer to the viewer, which ultimately helps to improve viewer experiences, even across devices. And for the video service provider, CDNs offer several operational benefits, as they deliver more intelligent traffic routing and load-balancing and offer better isolation and faster diagnosis of network issues.
CDNs capture logs for each segment of video sent to an end user, typically every few seconds for each stream. A powerful analytics system, along with the distributed nature of a CDN, allows the service provider to rapidly identify pattern anomalies, or isolate areas generating errors or congestion. One can identify whether an issue is due to a unique device type, an equipment failure, a geographical region, a particular type of traffic, or even a specific ABR format or content title for a single user. By rapidly processing analytics and generating appropriate alerts the provider can address issues quickly before they worsen—or last for a significant duration of a time-sensitive event.
Furthermore, while advanced CDN analytics at the server, network and application level can facilitate efficient and stable delivery, video habit insights derived from AI and ML can assist with customer retention by providing valuable insights at the subscriber level—such as predicting future actions based on past ones.
The Cisco Visual Networking Index found that by 2022 IP video traffic will be 82% of all video traffic. More than ever, finding the right technologies and expertise to help avoid network congestion must be a priority for video service providers. The uniqueness of consumption by viewer/household ensures that a "one size fits all" solution will not work when consumers demand a premium QoE. Analyzing what subscribers are looking for, tackling the bandwidth challenge without sacrificing quality and leveraging advanced technologies to improve compression efficiency are all key components to the equation that can result in both customer retention and acquisition.
And while planning for the unexpected is never easy, when video service providers are assured they have the right technology and expertise to quickly and accurately deliver more control and visibility, their subscribers are the ultimate winners.
Verizon's Darren Lepke, fuboTV's Geir Magnusson, and Reality Software's Nadine Krefetz discuss the present and future of OTT in this clip from Streaming Media East Connect 2020.
SVOD accounted for the lion's share of revenue—$12 billion or 58%